Last week I traveled to Singapore to give an invited talk at the AI4X conference, an exciting new venue gathering computer scientists, physicists and engineers to discuss how AI will accelerate scientific discoveries. My talk was the second of the opening day in the plenary session. I spent my 30 minutes arguing that fundamental science now needs another phase transition. After the embrace of machine learning for data analysis around 2012, we should allow AI tools to guide the design of new experiments, treating it as a co-design problem in which hardware and software are optimized together.
Anyway, I want to talk about something else here.
On my second day in Singapore, after the morning session of AI4X, I jumped in a taxi and was driven to Forest City, in nearby Malaysia. I had been invited there by members of the “Network School”, which I had initially mistaken for one of the typical PhD schools I give lectures to these days (e.g., in two days I will be doing precisely that in Orsay, at the AIPHY school -that one is a real PhD school). But no, more than a school lecture we should consider this as a seminar, as the “Network School” is something quite different. But let me tell you about it from the perspective of a first-time visitor.
The taxi ride crossed the Singapore-Malaysia border via a long bridge, and a few kilometres of authentic equatorial jungle. On the right, in the distance, I could see many structures that looked a lot like oil extraction facilities. And then Forest City appeared, in the guise of an astounding wall of tall, ugly cement buildings that could only have been built by Chinese architects. The effect was very similar to that used in several dystopian movies, where the city is a hyper-populated fortress and its cement boundaries separate it from nothingness. My impulse was to stop the driver to take a picture, but I repressed it – these days, who needs to take a picture of some striking landscape? However, it turns out it is not easy to find online the kind of picture I wanted to take. The best approximation I found by googling around is shown below.
Forest City, Johor Bahru, Malaysia. Marielle Descalsota/Insider, low resolution copy under fair use.
As I later learned, Forest city sits on land reclaimed from the sea. It is a 100-billion-$ Chinese project that failed, although not completely. Largely, those imposing buildings are empty, but there is a small population that indeed lives there. The driver stopped the car at what appeared to be a large hotel by the beach, and I was led in a large hall with a wide cafeteria-like open space, brimming with young people, most of them working on their laptops.
As my host had not arrived yet, I waded through the tables and found a robot ready to play a game of chess on a physical chessboard. I could not help starting a game, which unexpectedly ended badly for me. Still, I was excited to play with real pieces against a machine. The feeling of observing the robotic arm slowly picking up a captured piece, moving it to a specific spot outside the board, and then moving its capturing piece on the same square -the whole operation taking something like 10 seconds- was a bit anti-climatic, though: it made me remember that in the past 20 years AI has made much more progress with language than with physical interfaces.
When my host arrived, he first brought me to grab some food at a buffet, and during the quick lunch, he explained what the “Network School” really is, and what is the reason of its very strange location.
The network state
The network school is a community of about 500 (but growing) entrepreneurs, digital nomads, and other individuals who believe in the value of open-source software and share some other societal ideas and values, and who decided to physically move to live in the same place. The idea follows from a book titled “The network state”, where the author, Balaji Srinivasan, observes how many communities today arise in the cloud – in social media circles, e.g., where people easily aggregate in virtual groups. He argues that it would make sense for such groups to find a common physical space to live, such that the sharing of ideas, ideals, and work activities such as startups chould be amplified. Eventually, such a community could evolve by creating a governance structure and become semi-independent, depending on the local agreements with the hosting country.
The idea of a network state appears to invert the historical order in which societies normally emerge. Traditionally, states are built around geography: people happen to live in the same place, develop economic ties, and only later acquire a common identity and system of governance. The Network State proposes the opposite sequence. A community first forms online, around shared values, interests, and projects. Only later does it seek a physical location where its members can gather. Territory becomes the consequence of identity rather than its cause. Whether this inversion can work at scale is an open question, but it is undeniably an interesting social experiment.
Also, initially I found the idea behind the network school puzzling. If your goal is to build a startup, why move to a semi-isolated community? Wouldn't you want to be embedded in a large city? But after spending a few hours there I began to understand the attraction. The objective is not isolation; it is concentration. Universities, research laboratories, and technology clusters all derive part of their success from bringing together people who are ambitious, highly motivated, and willing to exchange ideas constantly. The Network School is trying to recreate that effect, but starting from a community selected by values rather than by profession or institutional affiliation.
The comparison with a university campus came naturally. Like a campus, the place concentrates young, ambitious people who are building projects and exchanging ideas. Unlike a university, however, there is no institution at the center. The community itself is the institution. Whether such a structure can remain cohesive over decades, survive internal disagreements, or resist the inevitable tension between collective ideals and individual incentives remains to be seen. But it is difficult not to admire the willingness of its members to test these questions in practice rather than merely discuss them online.
My host and I discussed the idea of how open-sourcing could be a way to fight the dominance of the super-empires of software we have today in the world, and the possibility that China may come to dominate key areas of AI infrastructure. I thought the goal was laudable, but in this game the winner will largely be determined by who has quicker, cheaper access to huge energy sources, so it remains unclear how an open source push can compete in the long term. I also questioned the stability of a group of people that shares values but is always threatened by the betrayal of members who can’t resist the call of going for-profit.
An engaged audience
After the interesting lunch discussions, we moved to the conference room, where I finally had the floor and delivered my lecture. I had wisely made it rather entry-level in scope, at least at the beginning (I spent some 15 minutes to explain what it is that we do in fundamental physics these days, what are the tools, the theoretical ideas, and the open questions). This was a good idea because a large part of the 25 or so attendees were not familiar with particle physics. But I was really surprised by their level of focus and engagement, and by the hour-long question-and-answer session that followed the one-hour lecture. Certainly among the best audiences I had in seminars and public lectures in many years (I gave over 90 such events in 30 years of practice).
Above: a post-mortem picture with some of the attendees. Provided by the author.
Waiting for the taxi that would return me to Singapore after the event, there was more time to discuss the network school and the environment. I got to learn how the place was put on the map, how the original project (four islands) was never completed (only one island exists). As it was explained to me, one of the selling points was that buyers of real estate in China could receive or obtain access to an apartment in Forest City. Having a place to retire to, or to spend vacations at just a few hours of flight from Beijing, could be attractive, and indeed there still are temporary or full residents that have continued to use their property in Forest City. But now, a few of those apartments are used by members of the Network School. In Forest City they have found a thriving, young community that shares their vision on the future of humanity; they have found some idea to fight for; and they are close to any business they want to run in Singapore anyway.
I also got to hear the issues they face in living there. While not exactly a ghost town, Forest City lacks a true hospital, and transportation links are not ideal. The weather in the area is also not precisely pleasant – it is very hot and humid, as in Singapore, so that people prefer to spend their time indoors. It remains to be seen whether the Network School will eventually evolve into the kind of self-governing entity envisioned by its founders, or whether it will remain a small but vibrant community embedded within Malaysia. Either way, it offers a fascinating glimpse of a possibility that until recently would have sounded like science fiction: that communities born on the internet might one day acquire a geography of their own.
